A statistical consultant was retained by a linen supplier to analyze a survey of hospital purchasing managers. After looking at the data, she realized that the survey had missed several key geographic areas and included some that were outside the target region. Some survey questions were ambiguous. Some respondents failed to answer all the questions or gave silly replies (one manager said he worked 40 hours a day). Of the 1,000 surveys mailed, only 80 were returned.
- What alternatives are available to the researcher?
- Might an imperfect analysis be better than none?
- If you were the consultant, how might you respond to the supplier?
Rarely is it the case that a statistical survey captures all the relevant factors. The researcher might want to create another survey with improved questions, fixing the geographical problem for example. In the case of incomplete or silly questions, the researcher could simply eliminate these results from the analysis. It may be the case that 50 or 60 of the 80 surveys contain "good" data. If ...
Descriptive statistics consultants are examined.